Data Visualisation

# > 🎨 > data visualisation with ggplot2
### Danielle Navarro

---

layout: true
 
<div class="my-footer">

</div>

---

## https://rstudio.cloud/project/901338

???

demo sequence:
- explain RStudio layout
- explain boring R text
- ctrl-L to clear boring text (verbalise shortcut)
- `library(tidyverse)`, explain messages
- `mpg`, explain data
- `?mpg`, explain "help" docs
- clear screen again
- `ggplot(data = mpg)`, explain each component while typing (use the word "function" early!)
- discuss output (note the popups and explain them)
- up arrow to recall last command (verbalise it)
- `+ geom_point(mapping = aes(x = displ, y = hwy))`
- talk them through this bit by bit, but "defer" explanation of `aes()` for now

---

---

???

- this is recap to solidify learning
- `ggplot()` creates the (still blank) plot using `data = mpg`

---

---

???

- then `geom_point()` adds the dots
- ... using the `mapping` (currently referred to as "blah")

---

???

- the `mapping` tells R *how* to connect the plot to the data
- in this case we map `x` onto the `displ` variable
- and we map `y` onto the `hwy` variable

---

???

- `displ` is essentially a measure of engine size
- `hwy` is fuel efficiency for highway driving

---

```r
library(tidyverse)
ggplot(data = mpg) +
geom_point(
  mapping = aes(
    x = displ,
    y = hwy
  )
) 
```
]

![](index_files/figure-html/hello-r-output-1.png)
]

---

## .hand[exercise 1...👯‍♀️]

- how to show the city mileage `cty` as the y-value?
- how to show the manufacture `year` as the x-value?  
- how to show the number of cylinders `cyl` as the colour?
- (bonus): how to include a regression line instead of dot points?

???

- they're not expected to know all the answers!
- questions 1-3 promote intuitive grasp of mappings
- question 4 is there to foreshadow geoms

---

.hand[i am very good at art]
<img src="img/My_First_Plot.gif" width="500" height="500" style="display: block; margin: auto;" />

---

.pull-left-narrow[
 .huge-blue-number[2]
]
.pull-right-wide[
 .larger[
 &nbsp; aesthetic 
 &nbsp; mappings 
 &nbsp; 🎨 🗺
 ]
]

---

---

???

- let's unpack the contents of that "blah", shall we?
- the `ggplot(data = mpg)` part tells us the `mpg` data go with our plot

---

???

- the `mpg` data contains "variables"
- here are some of those variables in the data...
- but how do they "map" to the plot???

---

???

- the visual content/features of a plot are "aesthetics" 
- `x` coordinates and `y` coordinates are aesthetics
- can include things like `size` and `colour` also
- (mention US/UK spelling)

---

???

- that "blah" was an **aesthetic mapping**
- it tells R which plot aesthetics are mapped onto which variables in the data

---

---

---

```r
ggplot(data = mpg) +
geom_point(
* mapping = aes(
*   x = displ,
*   y = hwy
* )
) 
```
]

---

```r
aes(
  x = displ,
  y = hwy 
) 
```
]
--
.pull-right[
.hand[.]

```
## Aesthetic mapping: 
## * `x` -> `displ`
## * `y` -> `hwy`
```
]

---

```r
ggplot(data = mpg) +
geom_point(
  mapping = aes(
    x = displ,
    y = hwy
  )
) 
```
]

![](index_files/figure-html/hello-r-output-2-1.png)
]

???

- ask them to set colour to the number of cylinders `cyl`
- (not a class exercise, just to segue...)

---

```r
ggplot(data = mpg) +
geom_point(
  mapping = aes(
    x = displ,
    y = hwy,
*   color = cyl
  )
) 
```
]

![](index_files/figure-html/unnamed-chunk-16-1.png)
]

???

- prompt: is something wrong here?
- hint: can you have 5.21 cylinders in an engine?
- (don't give them the answer yet...)

---

## .hand[exercise 2...👩‍💻]

code along with me while we...
- look at the `mpg` data
- recreate the last plot
- change the mapped **variables** (type `mpg` to see variables)
- change the mapped **aesthetics** (e.g. `x`, `y`, `color`, `size`, `fill`, `shape`)

???

- do the typing in rstudio cloud and get them to type along 
- get them to suggest possible code edits...
- let them discover the "continuous variable cannot be mapped..." error
- talk them through it and show them `factor()` as the answer

```{=html}
<div class="countdown" id="timer_606c4660" style="right:0;bottom:0;" data-warnwhen="0">
<code class="countdown-time">02:00</code>
</div>
```

---

.pull-left-narrow[
 .huge-blue-number[3]
]
.pull-right-wide[
 .larger[
 &nbsp; layered 
 &nbsp; plots 
 &nbsp; 🍰
 ]
]

---

---

---

---

---

```r
ggplot(data = mpg) +
*geom_point(
  mapping = aes(
    x = displ,
    y = hwy,
    color = cyl 
  )
) 
```
]

![](index_files/figure-html/unnamed-chunk-21-1.png)
]

---

```r
ggplot(data = mpg) +
geom_point(
  mapping = aes(
    x = displ,
    y = hwy,
    color = cyl 
  )
) + 
*geom_smooth(
* mapping = aes(
*   x = displ,
*   y = hwy
* )
*) 
```

```
## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
```
]

![](index_files/figure-html/unnamed-chunk-22-1.png)
]

---

## .hand[exercise 3...👩‍💻]

code along with me while we...
- recreate the plot with `geom_point()` only
- add the `geom_smooth()` layer
- add a third layer with `geom_rug()`

---

.pull-left-narrow[
  .huge-blue-number[4]
]
.pull-right-wide[
  .larger[
  &nbsp;&nbsp;tidy code
  ]
]

---

```r
ggplot(data = mpg) +
geom_point(
* mapping = aes(
*   x = displ,
*   y = hwy,
    color = cyl 
  )
) + 
geom_smooth(
* mapping = aes(
*   x = displ,
*   y = hwy
  ) 
) 
```
]

.pull-right[
.hand[can we eliminate this redundancy?]
- `x` and `y` values are the same
- make them **global** aesthetics
]

---

```r
ggplot(data = mpg,
* mapping = aes(
*   x = displ,
*   y = hwy
* )
) +
geom_point(
  mapping = aes(
    color = cyl 
  )
) + 
geom_smooth() 
```
]

---

```r
ggplot(data = mpg,
  mapping = aes( 
    x = displ, 
    y = hwy 
  ) 
) +
*geom_point(
  mapping = aes(
    color = cyl 
  )
) + 
*geom_smooth()
```
]

.pull-right[
 
- the `mapping` within the `ggplot()` command is "global"
- both "geoms" will use the `x` and `y` values that they "inherit"
]

---

```r
ggplot(data = mpg,
  mapping = aes( 
    x = displ, 
    y = hwy 
  ) 
) +
geom_point(
* mapping = aes(
*   color = cyl
* )
) + 
geom_smooth()
```
]

.pull-right[
 
- the `mapping` within the `ggplot()` command is "global"
- both "geoms" will use the `x` and `y` values that they "inherit"
- the `color` aesthetic remains "local" to `geom_point()`
]

---

```r
ggplot(
  data = mpg,
  mapping = aes(x = displ, y = hwy)
) +
geom_point(mapping = aes(color = cyl)) + 
geom_smooth() 
```
]

---

```r
ggplot(
  data = mpg,
  mapping = aes(x = displ, y = hwy)
) +
geom_point(mapping = aes(color = cyl)) + 
geom_smooth() 
```
]

.left[
- what are the **variables**?
- what are the **geoms**, and what do they do?
- identify the R **functions** this code uses 
- why did I say this is the "same" code as before?
]

---

---

```r
library(tidyverse)

dino <- read_csv("data_dino.csv")
picture <- ggplot(data = dino) + 
 geom_point(
 mapping = aes(
 x = horizontal, 
 y = vertical
 )
 )

plot(picture)
```

]

---

.pull-left-narrow[
 .huge-blue-number[6]
]
.pull-right-wide[
 .larger[
 &nbsp;naming &nbsp;things 
 ]
]

---

```r
dino <- read_csv("data_dino.csv")
*picture <- ggplot(data = dino) +
 geom_point(
 mapping = aes(
 x = horizontal, 
 y = vertical
 )
 )
plot(picture)
```
]

---

```r
dino <- read_csv("data_dino.csv")
*picture <- ggplot(dino) +
 geom_point(
 mapping = aes(
 x = horizontal, 
 y = vertical
 )
 )
plot(picture)
```
]

---

```r
*dino <- read_csv("data_dino.csv")
*picture <- ggplot(dino) +
 geom_point(
 mapping = aes(
 x = horizontal, 
 y = vertical
 )
 )
*plot(picture)
```
]

---

```r
dino <- read_csv("data_dino.csv") 
picture <- ggplot(dino) + 
 geom_point( 
* mapping = aes(
 x = horizontal,
 y = vertical
 ) 
 ) 
plot(picture) 
```
]

---

```r
dino <- read_csv("data_dino.csv")
picture <- ggplot(dino) + 
 geom_point(
* aes(
 x = horizontal, 
 y = vertical
 )
 ) 
plot(picture)
```

]

---

```r
dino <- read_csv("data_dino.csv")
picture <- ggplot(dino) + 
 geom_point(
 aes( 
* x = horizontal,
* y = vertical
 )
 ) 
plot(picture)
```

]

---

```r
dino <- read_csv("data_dino.csv")
picture <- ggplot(dino) + 
 geom_point(
 aes( 
* horizontal,
* vertical
 )
 ) 
plot(picture)
```

]

---

```r
dino <- read_csv("data_dino.csv")
picture <- ggplot(dino) + 
* geom_point(aes(horizontal, vertical))

plot(picture)
```

]

---

```r
dino <- read_csv("data_dino.csv")
*picture <- ggplot(dino) +
* geom_point(aes(horizontal, vertical))

plot(picture)
```

]

---

---

---

.pull-left[
<img src="img/handwriting.jpg" width="600" />
]
.pull-right[
- Martire, Growns & Navarro (2018) *What do the experts know? Calibration, precision, and the wisdom of crowds among forensic handwriting experts*
- https://psyarxiv.com/pq7kz
]

---

---

```r
ggplot(data = forensic) + 
 geom_boxplot(
 mapping = aes(
 x = band, 
 y = est
 )
 )
```
]
.pull-right[
 
![](index_files/figure-html/unnamed-chunk-36-1.png)
]

---

```r
ggplot(data = forensic) + 
* geom_violin(
 mapping = aes(
 x = band, 
 y = est
 )
 )
```
]
.pull-right[
 
![](index_files/figure-html/unnamed-chunk-37-1.png)
]

---

.pull-left-narrow[
 .huge-blue-number[8]
]
.pull-right-wide[
 .larger[
 &nbsp; facets 
 &nbsp; 🔲 🔲 
 &nbsp; 🔲 🔲 
 ]
]

---

```r
ggplot(forensic) + 
  geom_boxplot(aes(x = band, y = est)) 
```

---

```r
ggplot(forensic) + 
  geom_boxplot(aes(x = band, y = est)) + 
* facet_wrap(vars(handwriting_expert))
```

---

.pull-left-narrow[
 .huge-blue-number[9]
]
.pull-right-wide[
 .larger[
 &nbsp; glamour 
 &nbsp; graphics
 ]
]

---

<blockquote class="twitter-tweet tw-align-center">The recording of my <a href="https://twitter.com/hashtag/rstudioconf?src=hash&amp;ref_src=twsrc%5Etfw">#rstudioconf</a> talk, The Glamour of Graphics is up! 💅 Enjoy, and keep in mind these are guidelines, not rules. <a href="https://twitter.com/hashtag/rstats?src=hash&amp;ref_src=twsrc%5Etfw">#rstats</a> <a href="https://t.co/VHz05yAs09">https://t.co/VHz05yAs09</a>&mdash; Will Chase (@W_R_Chase) <a href="https://twitter.com/W_R_Chase/status/1228683209422057479?ref_src=twsrc%5Etfw">February 15, 2020</a></blockquote>

---

```r
pic <- ggplot(
 data = forensic
) + 
geom_boxplot(
 mapping = aes(
 x = band, 
 y = est, 
 fill = band
 )
) + 
facet_wrap(
 vars(handwriting_expert)
)
plot(pic)
```
]

---

```r
pic + 
 theme_minimal() 
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-39-1.png" width="600" height="300" />
]

---

```r
pic + 
 theme_minimal() + 
 scale_x_discrete(
 name = NULL,
 labels = NULL
 )
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-40-1.png" width="600" height="300" />
]

---

```r
pic <- pic + 
 theme_minimal() + 
 scale_x_discrete(
 name = NULL,
 labels = NULL
 ) + 
 scale_y_continuous(
 name = NULL
 )
pic
```
]

---

```r
pic <- pic + ggtitle(
 "Estimated frequency of handwriting features"
)
pic
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-42-1.png" width="600" height="300" />
]

---

```r
pic + 
 scale_fill_viridis_d()
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-43-1.png" width="600" height="300" />
]

---

```r
pic + 
 scale_fill_viridis_d(
 alpha = .5
 )
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-44-1.png" width="600" height="300" />
]

---

```r
pic + 
 scale_fill_viridis_d(
 alpha = .5,
 name = NULL
 )
```
]
.pull-right[
<img src="index_files/figure-html/unnamed-chunk-45-1.png" width="600" height="300" />
]

---

---

---

*Refresher?* ...try this lovely tutorial! 
https://ggplot2tutor.com/viz-pages/beginner_tutorial/beginner_tutorial/

*A classic?* ...look into "R for Data Science" 
https://r4ds.had.co.nz/data-visualisation.html

*This but extra?* ...I wrote an advanced workshop 
https://djnavarro.github.io/satrdayjoburg/

*A full book?* ...Keiran Healy's book is very good! 
https://socviz.co/